Research on Image Classification Based on Deep Learning

With the continuous exploration and research of researchers in the field of deep learning, deep learning has developed by leaps and bounds. Compared with traditional machine learning technology, deep learning has great advantages. The traditional machine learning algorithm for image classification needs to extract local features by hand. The emergence of deep learning has changed this situation and greatly promoted the development of image classification in the field of computer vision. The main content of this paper is to design a simple convolutional neural network model to classify three different common datasets. Comparative experiments are conducted by changing experimental parameters (such as activation function, pooling mode, size of output_size, etc.), and then the influence of different parameters on classification and recognition accuracy is analyzed under different data sets. Experimental results show that relu activation function and maximum pooling are more suitable for the classification of image data sets.